Results 131 to 140 of about 379,828 (307)
Scaling Law for Quantization-Aware Training [PDF]
arXivLarge language models (LLMs) demand substantial computational and memory resources, creating deployment challenges. Quantization-aware training (QAT) addresses these challenges by reducing model precision while maintaining performance. However, the scaling behavior of QAT, especially at 4-bit precision (W4A4), is not well understood.
arxiv
A Method of Approximate Second Quantization in the Theory of Superconductivity [PDF]
, 1960 Z.M. Galasiewicz
openalex +1 more source
Advanced Intelligent Systems, EarlyView.Recent advancements in wearable healthcare have brought accessible continuous glucose monitoring systems (CGMs) for diabetes management. To address the limitations of CGMs, closed‐loop systems utilizing monitored glucose levels for insulin dosing are being developed.
Wei Huang +5 more
wiley +1 more source
Energy‐Efficient Online Training with In Situ Parallel Computing on Electrochemical Memory Arrays
Advanced Intelligent Systems, EarlyView.By leveraging the intrinsic functionalities of electrochemical random‐access memory, the conductance response to pulse amplitude and quantity enables stochastic multiplication and parallel outer‐product operations between two vectors. This approach significantly accelerates weight gradient computations while reducing time complexity, latency, and ...
Yingming Lu +7 more
wiley +1 more source
Effects of quantization noise in digital filters [PDF]
, 1966 Bernard Gold, Charles M. Rader
openalex +1 more source
Voltage‐Summation‐Based Compute‐in‐Memory Technology with Capacitive Synaptic Devices
Advanced Intelligent Systems, EarlyView.Compute‐in‐memory (CIM) technologies leveraging capacitive coupling offer significant advantages in energy efficiency and IR‐drop elimination. This work introduces voltage‐summation‐based CIM technology, employing capacitive synaptic devices for matrix–vector multiplication.
Jung Nam Kim +8 more
wiley +1 more source
EURASIP Journal on Advances in Signal Processing, 2003 A dynamic programming-based optimization strategy for a temporal decomposition (TD) model of speech and its application to low-rate speech coding in storage and broadcasting is presented.
Bradley Alan B +2 more
doaj +1 more source
Configurable Kernel Map Implementation in Memristor Crossbar for Convolution Neural Network
Advanced Intelligent Systems, EarlyView.A configurable kernel map implementation using a memristor crossbar array is presented. The crossbar array area can be configured based on the number of read cycles per inference, which directly affects the inference speed. The algorithm underlying this scheme is described, and convolutional neural network operations are experimentally validated using ...
Gyeonghae Kim +3 more
wiley +1 more source